Circuit performance modeling by means of fuzzy logic

This paper presents a fuzzy-logic based approach to circuit performance modeling. Starting from a set of input-output vectors obtained by simulation (the training set), an automatic process adapts model parameters by means of neurofuzzy techniques. The resulting model can accurately reproduce the training set, even for a moderate number of rules. The results obtained in several analog and digital examples, show that accurate models can be obtained with a simple, completely automatic process. To demonstrate the usefulness of the proposed model, it is used to evaluate the circuit performances in the optimization process of an analog cell, achieving considerable speed-up